There are new weather sources available as of a month or so, that have a big impact on marine weather of US inland waters across the country. Namely the HRRR model forecasts in grib format that we have discussed recently for waters of the Pacific Northwest. These data are available nationwide, so I want to highlight a few other regions starting with the Great Lakes.
The Great Lakes already have excellent forecasting, but I think this new data will add to even that. More experienced users in that part of the world may know of better sources, but it seems to me one of the best is from the Great Lakes Coastal Forecast System (GLCFS). Below we take what we can get at this moment from that source. The GLCFS uses a model that is initiated every 12h, which makes it similar to the one at UW we discussed earlier.
Here is what we get going there now. [A video at the end illustrates the links.]
The Latest forecast as of 0400z is 4h old; the next surface wind report will be at 0628z, but that one (called Nowcast) does not have forecasts. The next forecast will be available at 1330z, which would be their 12z initialization. Thus it looks like it takes them 1h 30m to process the data and publish it, which is pretty fast.
Looking to other forecasts, we can compare that one to the free HRRR data online, from which at the moment we can get:
This data is 2h old. In other words, even though the model is run every hour, it takes about 2h to process and publish the results. This too is not bad. Global models take up to 3h for maps of the ocean.
The other option I want to highlight here is we can get this same HRRR data from the commercial service of ocens.com, which offers this data in grib format via their WeatherNet service. It costs about $3 to download 15h of wind and pressure over the Lakes region. An example is below where the data were loaded into Expedition. The Ocens grib files should load into most grib viewers or echart programs that display grib files.
This is the same data we viewed graphically above, so it is also 2h old, but viewed in grib format it is much more detailed. Furthermore, we at hand this way the full 15h of forecast, as shown in the video. We have more detail and we can read wind and pressure digitally at any cursor location. This can be compared to the free online view of the HRRR and GLCFS graphic data shown above.
For actual checks of the model accuracy, it seems there is one buoy right in the middle of the south end of the lake (Buoy 45007), but that one is not providing data at the moment, so we look to the Chicago area. This could be even more interesting region to look at as we see that this forecast calls for nominally 5 kts (±2.5) from the NW, which is notably different from farther offshore. Viewing the data in a grib viewer, we can read the actual forecast and do not have to rely on the definitions of these wind arrow symbols. We can find the actual wind in Chicago at 04z (turns out to be about now) from the NDBC site for this Chicago station CHII2, where we find that this particular forecast is more or less spot on. By the way, there is a really neat webcam showing conditions in this area.
There is every reason to believe that this HRRR forecast is the best available for the Great Lakes. It could be GLCFS and other sources of NWS Great Lakes forecasts are using it now, but it does not seem so from the above. Furthermore, even if they do use it for their 00z and 12z runs, the beauty of the HRRR model is it is run every hour with full assimilation of all observations over the past hour. In that sense it has to be by default superior to others for practical use.
Eventually we might have other sources of this data in grib format that will be available publicly at no charge. For now we thank Ocens for pioneering this presentation, which should help mariners across the country. Ocens does offer a free 3-day trial period for WeatherNet that can be used to download the data. We have a video on how to do this, which is good to check out when you have just 3 days to try it.
The video below illustrates the links above with a bit more annotation.